论文部分内容阅读
雷达目标识别中,目标一维距离像的分布常表现出明显的非线性和复杂性,这时经典的线性子空间方法的识别性能会有所下降.为此,本文提出非线性正则子空间法,通过对一维距离像进行非线性变换,使在原有空间线性不可分的一维距离像模式在高维空间有望具有线性可分性,从而提高目标的识别性能.对实测飞机数据的实验结果表明该方法的有效性.
In the radar target recognition, the distribution of the target one-dimensional distance images often shows obvious nonlinearity and complexity, then the recognition performance of the classical linear subspace method will decrease.Therefore, this paper proposes a nonlinear regular subspace method , The non-linear conversion of one-dimensional distance image can make the one-dimensional distance-insensitive linear distance model in the original space be linearly separable in the high-dimensional space, so as to improve the recognition performance of the target.Experimental results on measured aircraft data show that The effectiveness of this method.